Nick Jiang

Vision transformers have high-norm outliers that hurt performance and distort attention. While prior work removed them by retraining with “register” tokens, we find the mechanism behind outliers and make registers at ✨test-time✨—giving clean features and better performance! 🧵

So much of research centers on hope and faith that ppl who don’t take leaps of faith are unlikely to enjoy research

Updated paper! Our main new finding: by creating attention biases at test time—without extra tokens—we remove high-norm outliers and attention sinks in ViTs, while preserving zero-shot ImageNet performance. Maybe ViTs don’t need registers after all? x.com/nickhjiang/sta…

How do language models track mental states of each character in a story, often referred to as Theory of Mind? Our recent work takes a step in demystifing it by reverse engineering how Llama-3-70B-Instruct solves a simple belief tracking task, and surprisingly found that it…

Very interesting work. These outliers are the same outliers as in LLM.int8() and the attention sinks papers and suggest that outliers could be handled through test-time adaptations. LLMs are trickier due to natural registers (BOS token), but a similar approach might work.

This is really cool and useful vision work and will solve many of the problems I’ve been having.

One of the most interesting papers to read at the moment with extreme implications also for language transformers.

Artifacts in your attention maps? Forgot to train with registers? Use 𝙩𝙚𝙨𝙩-𝙩𝙞𝙢𝙚 𝙧𝙚𝙜𝙞𝙨𝙩𝙚𝙧𝙨! We find a sparse set of activations set artifact positions. We can shift them anywhere ("Shifted") — even outside the image into an untrained token. Clean maps, no retrain.